[
https://issues.apache.org/jira/browse/HBASE-29180?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17934610#comment-17934610
]
Andrew Kyle Purtell commented on HBASE-29180:
---------------------------------------------
+1
UHE can be caused by transient issues but if it repeats it likely to be
persistent especially in a dynamic environment like k8s.
The default number of retries after HBASE-28638 is 5 which is sufficient to
distinguish between the cases.
> Apply fail-fast retry limit for UnknownHostException
> ----------------------------------------------------
>
> Key: HBASE-29180
> URL: https://issues.apache.org/jira/browse/HBASE-29180
> Project: HBase
> Issue Type: Sub-task
> Affects Versions: 2.5.11
> Reporter: Viraj Jasani
> Assignee: Viraj Jasani
> Priority: Major
> Fix For: 2.7.0, 3.0.0-beta-2, 2.6.3
>
>
> As part of HBASE-28638, fail-fast retry limit has been introduced for errors
> like CallQueueTooBigException, SaslException, ConnectionClosedException. This
> helps limit the num of retries that RSProcedureDispatcher has to perform
> while executing remote procedures. Since the region open/close fails on the
> remote server, we also trigger SCP for the target server.
> We recently came across UnknownHostException as another example of where the
> remote calls can get stuck forever:
> {code:java}
> WARN [RSProcedureDispatcher-pool-98034] procedure.RSProcedureDispatcher -
> request to rs1.xyz,60020,1739254267238 failed due to
> java.net.UnknownHostException: Call to address=rs1.xyz:60020 failed on local
> exception: java.net.UnknownHostException: rs1.xyz:60020 could not be
> resolved, try=2867, retrying... , request params: open_region {
> open_info {
> region {
> ...
> ... {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
